Eric Horvitz is a computer scientist and the Chief Scientific Officer at Microsoft. He is a prominent figure in the fields of artificial intelligence and human-computer interaction (HCI), with a career spanning over four decades. His research has significantly influenced how AI systems handle uncertainty, make decisions, and interact with people.
Horvitz's work bridges theoretical foundations and practical applications. He has contributed to the development of machine learning methods, particularly around probabilistic reasoning and decision theory. His focus extends beyond algorithms to the user experience, aiming to create systems that anticipate human needs and assist effectively.
Early Career and Education
Horvitz received his medical degree from Stanford University in 1980 aid, later earned a PhD in medical informatics from the same institution in 1986. This background uniquely positioned him to apply AI to complex domains like healthcare. During his doctoral studies, he developed an interest in how computers could support clinical decision-making.
After completing his PhD, Horvitz joined Xerox PARC in 1986, where he worked on pioneering AI and HCI projects. His time at PARC allowed him to explore novel interaction paradigms i, including systems that adapt to user behavior. He then moved to Microsoft in 1992, where he has remained for most of his career.
Contributions to AI Research
At Microsoft, Horvitz led research on neural networks and probabilistic graphical models. He is known for his work on decision-theoretic control of AI systems, which involves using utility calculations to guide their actions. This research laid groundwork for later developments in deep learning and generative AI.
Horvitz also made key contributions to the field of attention and interruption management. He studied how computer systems could decide when to alert users, balancing costs and benefits. His work on 'differential diagnosis' and 'Bayesian networks' helped shape early clinical decision-support systems.
He has published widely and holds numerous patents. His research on 'mixed-initiative' interaction, where both humans and machines take proactive roles, remains influential in HCI and AI design.
Leadership at Microsoft
Horvitz has held several senior roles at Microsoft, including Director of the Adaptive Systems and Interaction group. He was appointed Distinguished Scientist and later became the Chief Scientific Officer. In this capacity, he oversees the company's long-term research strategy and coordinates efforts across its various labs.
He has been instrumental in steering Microsoft's AI ethics and policy. He has advocated for responsible development and deployment of AI, emphasizing transparency, fairness, and safety. His leadership has helped guide the creation of tools like Azure AI services, which incorporate many of his research insights.
Under his guidance, Microsoft Research has collaborated with academic institutions such as MIT CSAIL and Carnegie Mellon University on advancing AI. He has also fostered partnerships with other industry players, including Google DeepMind and Anthropic, to address shared challenges.
Influence on AI Policy and Ethics
Beyond technical contributions, Horvitz is known for his work on the societal implications of AI. He has served on committees and advisory boards, helping to shape national and international guidelines. He has spoken frequently about the need for 'human-centered' AI, aligning with themes from the OpenAI and DeepMind communities.
He co-chaired the Association for the Advancement of Artificial Intelligence (AAAI) panel on AI and ethics in 2008-2009, which called for increased research on AI's long-term impacts. He has also been involved with the Computing Research Association and the National Academy of Sciences on related initiatives.
His perspective emphasizes that AI should augment human capabilities rather than replace them. This philosophy has influenced Microsoft's product development and its public stance on regulatory matters.
Legacy and Recognition
Horvitz has received numerous awards, including fellowship in the American Association for the Advancement of Science (AAAS) and the ACM. He was elected to the National Academy of Engineering in 2013, recognizing his contributions to AI and HCI. He has also been honored with the ACM-AAAI Allen Newell Award.
He remains an active researcher, publishing on topics like large language models and their societal implications. His insights into transformer architectures and their potential have helped Microsoft integrate such models into products like Bing and Office.
Horvitz's career exemplifies the integration of rigorous science with practical engineering. His work has paved the way for many of today's AI systems, influencing a generation of researchers at institutions like Stanford AI Lab and Berkeley AI Research. His ongoing role as Chief Scientific Officer ensures that his vision continues to shape the future of computing.
References
- Horvitz, E. (1986). "A computational model of clinical decision making." PhD thesis, University of Stanford.
- Horvitz, E. (2001). "Principles of mixed-initiative user interfaces." Proceedings of CHI.
- Microsoft Research biography, 2023.
Artificial intelligence, machine learning, human-computer-interaction